The purpose of this work is to show how machine translation can be integrated into professional translation environments using two possible workflows. In the first workflow we demonstrate the real-time, sentence-by-sentence use of both rule-based and statistical machine translation systems with translation memory programs. In the second workflow we present a way of applying machine translation to full translation projects beforehand. We also compare and discuss the efficiency of statistical and rule-based machine translation systems, and propose some ideas about how these systems could be combined with translation memory technologies into a unified translation application.